Configure a rig →
NVIDIAConsumerGTX 10

GTX 1080 Ti for local AI

GTX 1080 Ti provides 11 GB of VRAM for local AI. In the LocalIA catalog, 164 out of 242 models run comfortably on a single card.

View all compatible models →Rig around the Ti ↗
VRAM
11GB
Category
Consumer
Series
GTX 10
Vendor
NVIDIA

Models that run comfortably

These models fit in 11 GB with room for context and stable inference.

★Qwen 2.5 14Bqwen8.8 GBcomfortableQ4 · / 11 GB
Qwen 2.5 Coder 14Bqwen8.8 GBcomfortableQ4 · / 11 GB
★Qwen 3 14Bqwen8.8 GBcomfortableQ4 · / 11 GB
★DeepSeek R1 Distill 14Bdeepseek8.8 GBcomfortableQ4 · / 11 GB
Phi-3 Medium 14Bphi8.8 GBcomfortableQ4 · / 11 GB
★Phi-4 14Bphi8.8 GBcomfortableQ4 · / 11 GB
★GLM-4.5 Airglm8.8 GBcomfortableQ4 · / 11 GB
★Qwen2.5 14B Instructqwen8.8 GBcomfortableQ4 · / 11 GB
★Qwen3 14Bqwen8.8 GBcomfortableQ4 · / 11 GB
★Qwen2.5 Coder 14B Instructqwen8.8 GBcomfortableQ4 · / 11 GB
★DeepSeek R1 Distill Qwen 14Bqwen8.8 GBcomfortableQ4 · / 11 GB
Llama 2 13Bllama8.2 GBcomfortableQ4 · / 11 GB
CodeLlama 13Bcodellama8.2 GBcomfortableQ4 · / 11 GB
OLMo 2 13Bolmo8.2 GBcomfortableQ4 · / 11 GB
Vicuna 13Bvicuna8.2 GBcomfortableQ4 · / 11 GB
★Mistral Nemo 12Bmistral9.2 GBcomfortableQ5 · / 11 GB
★Gemma 3 12Bgemma9.2 GBcomfortableQ5 · / 11 GB
StableLM 2 12Bstable9.2 GBcomfortableQ5 · / 11 GB
Solar 10.7Bsolar8.2 GBcomfortableQ5 · / 11 GB
Falcon 3 10Bfalcon7.7 GBcomfortableQ5 · / 11 GB
★Gemma 2 9Bgemma6.9 GBcomfortableQ5 · / 11 GB
Yi 1.5 9Byi6.9 GBcomfortableQ5 · / 11 GB
★Qwen 3.5 9Bqwen6.9 GBcomfortableQ5 · / 11 GB
★GLM-4 9Bglm6.9 GBcomfortableQ5 · / 11 GB
★GLM-4.7 Flashglm6.9 GBcomfortableQ5 · / 11 GB
GLM-4.1V 9B Thinkingglm6.9 GBcomfortableQ5 · / 11 GB
★NVIDIA Nemotron Nano 9Bnemotron6.9 GBcomfortableQ5 · / 11 GB
★gemma 2 9b itgemma6.9 GBcomfortableQ5 · / 11 GB
Llama 3 8Bllama8.9 GBcomfortableQ8 · / 11 GB
★Llama 3.1 8Bllama8.9 GBcomfortableQ8 · / 11 GB

Tight models

These models barely fit. They can run, but context and speed will be limited.

DeepSeek V2 Litedeepseek10.1 GBtightQ4 · / 11 GB
DeepSeek Coder V2 Litedeepseek10.1 GBtightQ4 · / 11 GB
StarCoder 2 15Bstarcoder9.4 GBtightQ4 · / 11 GB
★Phi-4 Reasoning Vision 15Bphi9.4 GBtightQ4 · / 11 GB

Unlocked in a 2x rig

With two cards in parallel (22 GB total), larger models become reachable.

CodeLlama 34Bcodellama21.4 GBtightQ4 · / 22 GB
Yi 1.5 34Byi21.4 GBtightQ4 · / 22 GB
★dolphin 2.9.1 yi 1.5 34byi21.4 GBtightQ4 · / 22 GB
★Qwen 2.5 32Bqwen20.1 GBtightQ4 · / 22 GB
★Qwen 2.5 Coder 32Bqwen20.1 GBtightQ4 · / 22 GB
★Qwen 3 32Bqwen20.1 GBtightQ4 · / 22 GB
★QwQ 32Bqwq20.1 GBtightQ4 · / 22 GB
★DeepSeek R1 Distill 32Bdeepseek20.1 GBtightQ4 · / 22 GB
Qwen 2.5 VL 32Bqwen20.1 GBtightQ4 · / 22 GB
★Granite 4 H-Small 32B-A9Bgranite20.1 GBtightQ4 · / 22 GB
GLM-4.6glm20.1 GBtightQ4 · / 22 GB
★GLM-4.7glm20.1 GBtightQ4 · / 22 GB
★GLM-5glm20.1 GBtightQ4 · / 22 GB
★GLM-5.1glm20.1 GBtightQ4 · / 22 GB
★Qwen3 32Bqwen20.1 GBtightQ4 · / 22 GB

Unlocked in a 4x rig

Server-style configuration (44 GB total) for the largest open-weight models.

★Mixtral 8x7Bmistral36.1 GBcomfortableQ5 · / 44 GB
Falcon 40Bfalcon30.7 GBcomfortableQ5 · / 44 GB
Command R 35Bcommand26.9 GBcomfortableQ5 · / 44 GB
Aya 23 35Baya26.9 GBcomfortableQ5 · / 44 GB

Similar GPUs

VRAM estimates updated 2026-05-12.